WebSphere Commerce EnterpriseLinuxFeature Pack 8WebSphere Commerce Version 7.0.0.9

Semi-automated search index replication

The production pattern implements a unique approach to search index replication that relies on file synchronization utilities and shared file systems.

The search index replication is handled in a manner very similar to the IBM HTTP Server file synchronization described in Synchronization of files to the Commerce IHS web server.

The search index is built on the primary WebSphere Commerce and Search node, which is on the Standalone part of the virtual system pattern. The file system that contains the index is shared with other nodes in the deployment by way of an NFS export. The search nodes in the pattern are subordinate nodes. They mount the NFS share to a local path under /mnt.

The following script is registered to run at startup: /opt/IBM/VSP/Commerce/CreateNodeSearch/mountDirSolr.sh. The script runs a scheduled job that uses rsync to pull changed index files from the master search node to the local subordinate node. The scheduled job sleeps for 6000 seconds between runs.